Characteristics prediction of sub-5 nm nanosheet field effect transistor (FET) using a machine learning approach

Due to the aggressive scaling down of logic semiconductors, the difficulty of semiconductor component processes has increased. As the structure of components becomes more complex, the time and cost of processes and simulations have risen. Machine learning is now being used to analyze the electrical characteristics data of semiconductor components and apply the trained machine learning to next-generation semiconductor development. Machine learning trained on process data and simulation results can quickly and accurately predict which electrical characteristics change significantly when the component’s structure changes and which parameters have a significant impact on the electrical characteristic changes. This paper presents suitable machine learning models for analyzing and predicting the electrical characteristics (on-current (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$I_{on}$ </tex-math></inline-formula>), off-current (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$I_{off}$ </tex-math></inline-formula>), threshold voltage (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$V_{th}$ </tex-math></inline-formula>), subthreshold swing (SS), and drain induced barrier lowering (DIBL)) and statistical distribution (mean and standard deviation of the electrical characteristics) resulting from geometrical variability (sheet thickness (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T_{wire}$ </tex-math></inline-formula>), sheet diameter (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D_{wire}$ </tex-math></inline-formula>), oxide thickness (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T_{ox}$ </tex-math></inline-formula>), gate length (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$L_{g}$ </tex-math></inline-formula>), spacer length (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$L_{sp}$ </tex-math></inline-formula>), gate metal work-function (WF)) in nanosheet field-effect transistor (NSFET), which are a next-generation logic device. Machine learning models, including regulation-based models (Ridge and LASSO) and tree-based models (decision tree (DT), random forest (RF), extreme gradient boost (XGBoost), and light gradient boost machine (LGBM)), are trained on technology computer-aided design (TCAD) simulation data. The LGBM more accurately predicts the electrical characteristics and statistical distribution of the NSFET than the other models. Additionally, we analyze the effect of geometrical variability on the NSFET based on feature importance.

In this paper, the junctionless Fin Field Effect Transistor (FinFET) and nanosheet Field Effect Transistor (NSFET) with a gate length of 12 nm are implemented using the Sentaurus Technology Computer-Aided Design (TCAD) tool. To compare the junctionless FinFET and NSFET, simulations are done at constant threshold voltage. The NSFET outperformed FinFET in terms of current driving capabilities, Subthreshold Swing (SS), Drain Induced Barrier Lowering (DIBL), and intrinsic voltage gain (AV). Further, the device design guidelines are presented for FinFET and NSFET in terms of geometrical parameters. The simulation indicates that downscaling the gate length from 16 to 8 nm leads to an increase in SS and DIBL by 21 and 68.49 % in FinFET whereas 19 and 70.14 % in nanosheet FET. The height variation of FinFET seems to make the least impact on short channel effects (SCEs) while scaling the thickness of NSFET from 9 to 5 nm improves the DIBL and SS by 61.9 % and 15.54 % respectively. In the case of scaling the width of FinFET from 10 to 5 nm, DIBL and SS increase by 55.4 % and 14 % whereas scaling of nanosheet width from 24 to 12 nm gives 19.44 % and 1.37 % improvement in DIBL and SS, respectively.

We investigated process variation effect of 3D NAND flash memory cell, especially about geometric variation using a machine learning (ML) model. Geometric variability sources impact on variation of device's electrical parameters such as threshold voltage $(\mathbf{V}_{\mathbf{t}})$ , subthreshold swing (SS), transconductance $(\mathbf{g}_{\mathbf{m}})$ and on-current $(\mathbf{I}_{\mathbf{on}})$ . All these data were analyzed with 3D stochastic Technology Computer-Aided Design (TCAD) simulation and trained through ML model, which is composed of artificial neural network (ANN). The model has multi-input and multi-output (MIMO) structure and deep hidden layers to train and predict complex data of process variation. In order to make ML model more accurate, simulation for constructing training data set was carried out with a large number of random unit cells, which are cut from various strings. The completed ML model was tested with random test data set which had not been used for training to prove its accuracy. Through the test process, ML model showed the error of up to 5% and proved the accuracy of prediction.

The scaling of nanosheet (NS) field effect transistors (FETs) from the 12 nm gate length to the ultimate gate length of 10 nm for sub-2 nm nodes brings additional technological challenges. Here, 3D finite element Monte Carlo simulations are employed to explore how to alter the NS architecture to increase the drive current (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${I}_{\mathrm {\mathbf { DD}}}$ </tex-math></inline-formula>) because the gate scaling to 10 nm results in a decline of the current (by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {10.7}$ </tex-math></inline-formula>%). <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${I}_{\mathrm {\mathbf {DD}}}$ </tex-math></inline-formula> of the 10 nm gate length NS FET will increase by 11% if the maximum n-type source/drain doping reaches <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1\times 10^{20} \mathrm {cm^{-3}}$ </tex-math></inline-formula>, or increase by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {3.8}$ </tex-math></inline-formula>% if the high-<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\kappa $ </tex-math></inline-formula> dielectric layer equivalent oxide thickness (EOT) is less than <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {1.0}$ </tex-math></inline-formula> nm. The reduction in the channel width below 40 nm or the reduction in the channel thickness below 5 nm will substantially decrease IDD. The sub-threshold figures of merit like the sub-threshold slope (SS) will decrease from 75 to 73 mV/dec, while the drain-induced barrier lowering (DIBL) will increase from 32 to 77 mV/V. Finally, the effect of strain to increase the drive current is strongly limited by quantum confinement. <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${I}_{\mathrm {\mathbf {DD}}}$ </tex-math></inline-formula> will increase by 3% and by 14% in the 10 nm gate NS FET with the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\langle 110\rangle $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\langle 100\rangle $ </tex-math></inline-formula> channel orientations, respectively, when a strain of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {0.5}$ </tex-math></inline-formula>% is applied to the channel, with a negligible increase for larger strain values (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {0.7}$ </tex-math></inline-formula>% and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathbf {1.0}$ </tex-math></inline-formula>%).

A machine learning (ML) model by combing two autoencoders and one linear regression model is proposed to avoid overfitting and to improve the accuracy of Technology Computer-Aided Design (TCAD)-augmented ML for semiconductor structural variation identification and inverse design, without using domain expertise. TCAD-augmented ML utilizes TCAD simulations to generate sufficient data for ML model development when experimental data are inadequate. The ML model can then be used to identify semiconductor structural variation for given experimental electrical measurements. In this study, the variation of layer thicknesses in the p-i-n diode is used as a demonstration. An ML model is developed to predict the diode layer thicknesses based on a given Current-Voltage (IV) curve. Although the variations of interest can be incorporated easily in TCAD simulations to generate ML training data, the TCAD-augmented ML model generally is overfitted and cannot predict the variations in experiment well due to hidden variables which also alters the IV curves. We show that by using an autoencoder, this problem can be solved. To verify the effectiveness, another set of TCAD simulation data is generated with hidden variables (dopant concentration variation) to emulate experimental data. Testing on the second set of data shows that the proposed model can avoid overfitting and has up to 15 times improvement in accuracy in thickness prediction. Moreover, this model is used successfully to perform inverse design and can capture an underlying physics that cannot be described by a simple physical parameter.

The continued downscaling of semiconductor devices to overcome the limitations of traditional FinFETs and mitigate short-channel effects (SCEs) necessitates the adoption of novel device architectures. The gate-all-around (GAA) nanosheet field-effect transistor (NSFET) has emerged as a promising candidate for sub-7-nm and beyond technology nodes due to its superior electrostatic control and improved current drive capability. This research presents a comprehensive modeling and performance evaluation of a 20 nm gate length (Lg) silicon-based NSFET using three-dimensional (3D) technology computer-aided design (TCAD) simulations. The study systematically investigates the influence of key geometrical parameters, such as nanosheet width (Wns) and thickness (Tns), on critical direct-current (DC) performance metrics, including: Ion, Ioff, SS, DIBL. The simulation results demonstrate that the 20 nm NSFET exhibits excellent performance DC characteristics, and a steep subthreshold swing, making it highly suitable for low-power and high-speed switching applications.

Machine learning (ML) is a subdomain of artificial intelligence that enables computers to abstract patterns from data without explicit programming. A myriad of impactful ML applications already exists in orthopaedics ranging from predicting infections after surgery to diagnostic imaging. However, no systematic reviews that we know of have compared, in particular, the performance of ML models with that of clinicians in musculoskeletal imaging to provide an up-to-date summary regarding the extent of applying ML to imaging diagnoses. By doing so, this review delves into where current ML developments stand in aiding orthopaedists in assessing musculoskeletal images. This systematic review aimed (1) to compare performance of ML models versus clinicians in detecting, differentiating, or classifying orthopaedic abnormalities on imaging by (A) accuracy, sensitivity, and specificity, (B) input features (for example, plain radiographs, MRI scans, ultrasound), (C) clinician specialties, and (2) to compare the performance of clinician-aided versus unaided ML models. A systematic review was performed in PubMed, Embase, and the Cochrane Library for studies published up to October 1, 2019, using synonyms for machine learning and all potential orthopaedic specialties. We included all studies that compared ML models head-to-head against clinicians in the binary detection of abnormalities in musculoskeletal images. After screening 6531 studies, we ultimately included 12 studies. We conducted quality assessment using the Methodological Index for Non-randomized Studies (MINORS) checklist. All 12 studies were of comparable quality, and they all clearly included six of the eight critical appraisal items (study aim, input feature, ground truth, ML versus human comparison, performance metric, and ML model description). This justified summarizing the findings in a quantitative form by calculating the median absolute improvement of the ML models compared with clinicians for the following metrics of performance: accuracy, sensitivity, and specificity. ML models provided, in aggregate, only very slight improvements in diagnostic accuracy and sensitivity compared with clinicians working alone and were on par in specificity (3% (interquartile range [IQR] -2.0% to 7.5%), 0.06% (IQR -0.03 to 0.14), and 0.00 (IQR -0.048 to 0.048), respectively). Inputs used by the ML models were plain radiographs (n = 8), MRI scans (n = 3), and ultrasound examinations (n = 1). Overall, ML models outperformed clinicians more when interpreting plain radiographs than when interpreting MRIs (17 of 34 and 3 of 16 performance comparisons, respectively). Orthopaedists and radiologists performed similarly to ML models, while ML models mostly outperformed other clinicians (outperformance in 7 of 19, 7 of 23, and 6 of 10 performance comparisons, respectively). Two studies evaluated the performance of clinicians aided and unaided by ML models; both demonstrated considerable improvements in ML-aided clinician performance by reporting a 47% decrease of misinterpretation rate (95% confidence interval [CI] 37 to 54; p < 0.001) and a mean increase in specificity of 0.048 (95% CI 0.029 to 0.068; p < 0.001) in detecting abnormalities on musculoskeletal images. At present, ML models have comparable performance to clinicians in assessing musculoskeletal images. ML models may enhance the performance of clinicians as a technical supplement rather than as a replacement for clinical intelligence. Future ML-related studies should emphasize how ML models can complement clinicians, instead of determining the overall superiority of one versus the other. This can be accomplished by improving transparent reporting, diminishing bias, determining the feasibility of implantation in the clinical setting, and appropriately tempering conclusions. Level III, diagnostic study.

In this paper, we present an artificial neural network (ANN)-based compact model to evaluate the characteristics of a nanosheet field-effect transistor (NSFET), which has been highlighted as a next-generation nano-device. To extract data reflecting the accurate physical characteristics of NSFETs, the Sentaurus TCAD (technology computer-aided design) simulator was used. The proposed ANN model accurately and efficiently predicts currents and capacitances of devices using the five proposed key geometric parameters and two voltage biases. A variety of experiments were carried out in order to create a powerful ANN-based compact model using a large amount of data up to the sub-3-nm node. In addition, the activation function, physics-augmented loss function, ANN structure, and preprocessing methods were used for effective and efficient ANN learning. The proposed model was implemented in Verilog-A. Both a global device model and a single-device model were developed, and their accuracy and speed were compared to those of the existing compact model. The proposed ANN-based compact model simulates device characteristics and circuit performances with high accuracy and speed. This is the first time that a machine learning (ML)-based compact model has been demonstrated to be several times faster than the existing compact model.

Machine learning (ML) and deep learning (DL) models can play an important role when it comes to modelling complicated processes. Such capability is necessary for hydrological and climate-related applications. Generally, ML models utilize precipitation and temperature time series of a basin as input to develop a lumped rainfall-runoff model to simulate streamflow at the basin outlet. However, when it is divided into several sub-basins, Graph Neural Networks (GNN) can consider each sub-basin as a node and link them together using a connectivity matrix to account for spatial variations of hydroclimatic variables. In this study, GNN and various ML models with different types of architecture, ranging from neural networks, tree-based structure, and gradient boosting, were exploited for daily streamflow simulation over different case studies. For each case study, the basin was divided into a few sub-basins for which daily precipitation and temperature data were aggregated and used as input. For training GNN, the connection matrix of sub-basins was also used as input. Basically, 75% of historical records were utilized to train GNN and different ML models, e.g., artificial neural networks, support vector machine, decision tree, random forest, eXtreme Gradient Boosting (XGBoost), Light Gradient-Boosting Machine (LightGBM), and Category Boosting (CatBoost), while the rest was used for testing. Streamflow simulation was conducted with/without considering seasonality impact and lag times. The obtained results clearly demonstrate that considering seasonality and time lags can enhance accuracy of streamflow predictions based on Kling–Gupta efficiency (KGE). Furthermore, GNN with seasonality impact and time lags achieved promising results across different case studies with KGE&gt;0.85 for training and KGE&gt;0.59 for testing data, respectively. Among ML models, boosting models, e.g., LightGBM and XGBoost, performed slightly better than other ML models. for Finally, this comparative analysis provides valuable insights for ML/DL applications in climate change impact assessments.Acknowledgements: This research work was carried out as part of the TRANSCEND project with funding received from the European Union Horizon Europe Research and Innovation Programme under Grant Agreement No. 10108411.

Nanosheet Field Effect Transistor (NSFET) is a viable contender for future scaling in sub-7-nm technology. This paper provides insights into the variations of DC FOMs for different geometrical configurations of the NSFET. In this script, the DC performance of 3D GAA NSFET is analyzed by varying the device's width and thickness. Moreover, the gate length is scaled from 20 nm to 5 nm to check for the device suitability for continuous scaling in logic applications. The thickness and width of each nanosheet are varied in the range of 5 to 9 nm and 10 to 50 nm, respectively, to analyze the performance dependency on the geometry of the device. The impact of geometry of NSFET on various DC performance metrics like transfer characteristics, sub-threshold swing (SS), on current (ION), off current (IOFF), switching ratio (ION/IOFF), threshold voltage (Vth), and drain induced barrier lowering (DIBL) are studied. On top of that, the device's electrical characteristics are analyzed for a wide range of temperatures from -43oC to 127oC to identify the temperature compensation point and is observed at VGS = 0.55 V and ID = 3.86 × 10-6 A. Furthermore, the vital process parameter, work function variations on transfer characteristics of the device is analyzed. Moreover, the analyses reveal that, for sub -7 nm, the NSFET is a potential device for high performance and suitable for logic applications.

Comprehensive performance enhancement of a negative-capacitance nanosheet field-effect transistor with a steep sub-threshold swing at the sub-5-nm node

Highlights Machine Learning (ML) models are identified, reviewed, and analyzed for HAB predictions. Data preprocessing is vital for efficient ML model development. ML models for toxin production and monitoring are limited. Abstract. Harmful algal blooms (HABs) are detrimental to livestock, humans, pets, the environment, and the global economy, which calls for a robust approach to their management. While process-based models can inform practitioners about HAB enabling conditions, they have inherent limitations in accurately predicting harmful algal blooms. To address these limitations, Machine Learning (ML) models can potentially leverage large volumes of IoT data to aid in near real-time predictions. ML models have evolved as efficient tools for understanding patterns and relationships between water quality parameters and HAB expansion. This review describes ML models currently used for predicting and forecasting HABs in freshwater ecosystems and presents model structures and their application for predicting algal parameters and related toxins. The review revealed that regression trees, random forest, Artificial Neural Network (ANN), Support Vector Regression (SVR), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) are the most frequently used models for HABs monitoring. This review shows ML models' prowess in identifying significant variables influencing algal growth, HAB drivers, and multistep HAB prediction. Hybrid models also improve the prediction of algal-related parameters through improved optimization techniques and variable selection algorithms. While ML models often focus on algal biomass prediction, few studies apply ML models for toxin monitoring and prediction. This limitation can be associated with a lack of high-frequency toxin datasets for model development, and exploring this domain is encouraged. This review serves as a guide for policymakers and researchers to implement ML models for HAB prediction and reveals the potential of ML models for decision support and early prediction for HAB management. Keywords: Cyanobacteria, Freshwater, Harmful algal blooms, Machine learning, Water quality.

This study focuses on using various machine learning (ML) models to evaluate the shear behaviors of ultra-high-performance concrete (UHPC) beams reinforced with glass fiber-reinforced polymer (GFRP) bars. The main objective of the study is to predict the shear strength of UHPC beams reinforced with GFRP bars using ML models. We use four different ML models: support vector machine (SVM), artificial neural network (ANN), random forest (R.F.), and extreme gradient boosting (XGBoost). The experimental database used in the study is acquired from various literature sources and comprises 54 test observations with 11 input features. These input features are likely parameters related to the composition, geometry, and properties of the UHPC beams and GFRP bars. To ensure the ML models' generalizability and scalability, random search methods are utilized to tune the hyperparameters of the algorithms. This tuning process helps improve the performance of the models when predicting the shear strength. The study uses the ACI318M-14 and Eurocode 2 standard building codes to predict the shear capacity behavior of GFRP bars-reinforced UHPC I-shaped beams. The ML models' predictions are compared to the results obtained from these building code standards. According to the findings, the XGBoost model demonstrates the highest predictive test performance among the investigated ML models. The study employs the SHAP (SHapley Additive exPlanations) analysis to assess the significance of each input parameter in the ML models' predictive capabilities. A Taylor diagram is used to statistically compare the accuracy of the ML models. This study concludes that ML models, particularly XGBoost, can effectively predict the shear capacity behavior of GFRP bars-reinforced UHPC I-shaped beams.

Machine learning approaches for formation matrix volume prediction from well logs: Insights and lessons learned

Source/drain (S/D) variations in sub-5-nm node fin and nanosheet field-effect transistors (NSFETs) were thoroughly analyzed by using fully calibrated technology computer-aided design (TCAD). S/D open and contact critical dimensions (OCD and CCD) vary during anisotropic etching for silicide and S/D epi formations, respectively, and these vary dc/ac performances. OCD varies S/D resistances and parasitic capacitances, but slight RC delay variations occur. CCD affects OFF-state currents ( ${I}_{{ \mathrm{\scriptscriptstyle OFF}}}$ ) and RC delay, but differently in terms of fin field-effect transistors (FinFETs) and NSFETs. As the S/D epi enlarges by CCD changes, high subfin leakage of FinFETs flows into the fin bottom regions, thus varying ${I}_{{ \mathrm{\scriptscriptstyle OFF}}}$ greatly. NSFETs have the dielectric layers beneath the S/D epi; therefore, the ${I}_{{ \mathrm{\scriptscriptstyle OFF}}}$ variations are smaller. Both effective currents ( ${I}_{{\text {eff}}}$ ) and gate capacitances ( ${C}_{{\text {gg}}}$ ) of FinFETs vary in the same direction, whereas the NSFETs have constant ${C}_{{\text {gg}}}$ with respect to CCD changes because of the tradeoff between the intrinsic and parasitic capacitances. Although this effect increases the RC delay variations of the NSFETs, p-type FinFETs have the largest RC delay variations due to the greatest relative variations of ${I}_{{\text {eff}}}$ . Thus, the NSFETs are much immune to CCD variations compared with FinFETs in terms of ${I}_{{ \mathrm{\scriptscriptstyle OFF}}}$ and RC delay.