Mobile QR Code QR CODE

  1. (Power Electronics R&D Team, LS Electric, Dongan-gu Anyang 14118, Korea)
  2. (Department of Electronic Engineering, Sogang University, 35 Baekboem-ro, Mapo-gu, Seoul 04107, Korea)
  3. (Department of Electronic Engineering, Myongji University, Yongin 17058, Korea)
  4. (Department of Electrical and Electronics Engineering, Konkuk University, Seoul 05029, Korea)
  5. (Department of Electronic Engineering, Ajou University, Suwon 16499, Korea)



Machine learning, TFET, tunneling, band to band tunneling

I. INTRODUCTION

Recently, Tunnel Field-Effect Transistors (TFETs) have been studied as promising alternatives to metal-oxide-semiconductor field-effect transistors (MOSFETs) [1-6], particularly for very low-power application. Based on carrier injection through band-to-band tunneling (BTBT), significant progress has been made in achieving a subthreshold swing (SS) of less than 60 mV/dec at room temperature (RT) and minimizing low-level off-state current (Ioff). Nevertheless, in contrast to MOSFETs, TFETs face a challenging issue with low on-state current (Ion). To address this, a high-${\kappa}$ / metal gate (HKMG) materials have been adopted instead of polysilicon gate. The HKMG, capable of reducing gate leakage and ensuring high channel controllability, has shown promise in improving TFET performance [7-10].

Despite advancements in Ion characteristics, Titanium Nitride (TiN), a common HKMG material, introduces work function variation (WFV). Sputtered TiN tends to crystallize predominantly in <200> (60%) and <111> (40%), corresponding to WFs 4.6 eV and 4.4 eV respectively. This non-uniformity in metal gates contributes to WFV, influencing TFET current variations [11-13].

Therefore, when implementing the TFETs into real complementary metal-oxide-semiconductor (CMOS) circuits, it is essential to examine electrical performance variations in relation to WFV. However, studies addressing this issue have been limited. Some researchers have proposed nanowire TFETs as the gate of nanowire TFET effectively reduces WFV by minimizing the affected channel area [14-16]. Despite efforts to improve this aspect, WFV still persists. Consequently, identifying the primary cause has become crucial for reducing or eliminating WFV [17].

The primary objective of this study is to establish causal relationships using a machine learning (ML) approach [18-20]. ML facilitates a comprehensive analysis of TFETs even in the absence of complete WFV data samples. It allows for predictions by adjusting output parameters and constructing a model that captures variations in electrical operations associated with WFV, combining the variation of current.

ML allows to identify the cause with a limited dataset. However, once relevant parameters are identified, ML models need optimization by pinpointing exceptional cases that were not part of the training data. Models, trained on less data tend to exhibit low R2 value for parameters with less relevance. Therefore, it was necessary to reduce sensitivity to identify parameters that have a direct impact.

Transistor samples have been verified using technology computer-aided design (TCAD) and simulations were conducted using Synopsys Sentauraus [20,21]. With relevant parameters, ML predicts exceptional situations by different WF in each area. In Section 2, we delve into the valuable output parameters (e.g., SS, Threshold voltage (Vth), minimum current (Imin), turn-on voltage (Von)) influenced by WFV. Section 3 outlines the most relevant input parameter identified by ML. Subsequently, in Section 4, the ML model is constructed using data highly dependent on Drain Voltage (Vd). Finally, Section 5 presents predictions for exceptional situations.

II. DEVICE DESIGN AND METHODOLOGY

In Fig. 1(a), a bird’s eye view of a nanowire TFET is presented. The simulation, conducted using TCAD, features a gate oxide thickness (Tox) of 1 nm with SiO2. Arsenic and boron are used as the dopant atoms for n-type and p-type doping, respectively. The structure of the TFET is p-i-n, with a TiN gate and a channel length of 20 nm. All simulations are performed at RT and the design parameters are summarized in Table 1. Fig. 1(b) displays a bird’s eye view of a WF randomized model, while Fig. 1(c) illustrates the Y-axis cross-section. The grain size of TiN is assumed to be an identical, forming a 5 ${\times}$ 5 nm2 of square shape [22].

Table 1. Parameters of structure

Parameters

Value

Source doping concentration (Ns)

Drain doping concentration (ND)

Body doping concentration (NB)

Gate work-function

Channel length (Lch)

Channel diameter (dch)

Average metal grain size

Gate oxide thickness (Tox)

Drain Voltage (VD)

1020 cm-3 (p-type)

1020 cm-3 (n-type)

1018 cm-3 (p-type)

Variable

20 nm

20 nm

5 × 5 nm2

1 nm

1 V

Fig. 1. (a) Bird’s eye view of nanowire TFET; (b) Structure having random WF on gate; (c) Y-axis cross section and divided gate area; (d) Z-axis cross section and divided gate area.
../../Resources/ieie/JSTS.2024.24.2.96/fig1.png

Fig. 1(d) shows Z-axis view, where the gate covering channel is divided into 8 areas that contact the oxide part. Given the 5 ${\times}$ 5 nm$^{2}$ square shape of the TiN grain, the gate area is further divided into 32 units. WFV for each gate area is randomly assigned, taking into account these probabilities.

A thousand structures were generated for ML, with 32 WF parameters extracted from each grain near the gate. Additionally, 16 parameters, including conduction band and valance band electron volt values at Vg = 0.5 V, Vg =1.5 V, were employed to describe the WFV profile in nanowire TFET in Fig. 2(a). The parameter mark points contact with the dotted line on the graph (0.025, 0.075, 0.125, 0.175 ${\mu}$m). In total, 48 parameters were used for input to train the ML model.

Fig. 2. (a) Energy band diagram atVg= 0.5 V andVg= 1.5 V; (b)Id-Vgcurve atVd= 1.0 V; (c) Summary of ML model; (d) Visualization of ML algorithm.
../../Resources/ieie/JSTS.2024.24.2.96/fig2.png

For the ouput, four parameters (SS, Vth, Ioff, Von) were extracted from the Id-Vg curve in Fig. 2(b). This curve illustates current’s variation with gate voltage. Imin, the minimum current flowing through the device, is measured at the lowest value of drain current, marked by green circle. SS, the subthreshold swing, represents the slope with increasing current and is measured at the point where the current increases 100 times from Imin, indicated by the purple circle. Von is the gate voltage value defining the on-state, measured when the drain current reaches 10-15 A/${\mu}$m marked on the blue line. Vth, the voltage at which the device operates, is measured when the drain current reaches 10-9 A/${\mu}$m indicated on the red line. Its distribution follows a Gaussian distribution due to randomized grain.

To design an ML algorithm and enable predictions, the process followed the four steps, as illustrated in Fig. 2(c):

Step 1: Split the data into training, testing, and validation sets.

Step 2: Construct an ML algorithm by training it with the designated data.

Step 3: Monitor the epoch and loss until the weights are optimized.

Step 4: Adjust or refine the neural network architecture as needed and deploy it accordingly.

Following the outlined procedure, thousands of data points were divided into the ratio of 8:1:1 for training, testing, and validation. The built-in ML algorithm utilized a dense layer structure of 48 - 32 - 16 - 4 [23]. MinMaxScaler was employed to normalize parameters, ensuring consistency with the formula (i.e., (X - Xmin) / (Xmax - Xmin)). Categorical Cross-entropy served as the loss function for each dense layer in Fig. 2(d), while the Adaptive Moment Estimation (ADAM) optimizer was chosen for accurate error correction, with a learning rate (LR) set to 0.001 [24]. Fig. 3(a) depicts the loss of the ML model during the building process. The appropriate epoch, indicating the point at which the loss reaches 0.04136, was determined through the observation of loss size as the epoch increased. For quantitative correlation verification, R2 values were extracted for various parameters. R2 values, a statistical measure of fit, were checked for each parameter—SS (0.7738), Vth (0.9933), Imin (0.5608), and Von (0.9938), as shown in Fig. 3(b).

Fig. 3. (a) Loss of ML model from each epoch; (b) R2value of each output parameters.
../../Resources/ieie/JSTS.2024.24.2.96/fig3.png

Comparing the R2 values for each parameter reveals a strong correlation between WFV and Vth,, Von, while the relationship with Imin and SS is less shown. Vth and Von exhibit higher R2 values than SS and Imin. With sufficient data, the model could approach an ideal state for predicting all parameters accurately.

III. FINDING THE DETAILED CAUSE

Building upon the findings in Section 2 regarding the association between WFV, Von, and Vth, mitigating the impact on WFV has become a key focus. In TFET, Von is influenced by a current at the point where BTBT is maximized. To identify a more valuable input parameter, it is essential to locate the area with maximized BTBT. In Fig. 1(c), the gate is segmented into four areas designated as gate 1, 2, 3, and 4 from the source (far right) of the structure. To assess the impact of each gate region on Von using ML, two gates were grouped together. These groups of the gates served as the input for the ML model, and evaluating the R2 value elucidates the correlation between Von and the group.

Sixteen parameters from the gates and eight parameters from the conduction and valence bands in the energy diagram were selected as input, while two parameters (Von, Vth) were designated as the output. The dense layer architecture was configured as 24 - 16 - 8 - 2, utilizing the ReLU function [25]. The ADAM optimizer was employed for precise error correction, with a learning rate set to 0.001.

The comparison of R2 values among different gate groups revealed that gate 1 and gate 2 exhibit a substantial correlation with Von, as depicted in Fig. 4(a). Notably, the group containing gate 2 demonstrated R2 values consistently at or above 0.9, indicating a robust correlation with Von. This outcome highlights the significant influence of gate 2 on Von. Considering that a significant portion of current in TFET is attributed to BTBT, as illustrated in Fig. 4(b) for an approximate gate voltage (Vg = 0.5 V) estimating Von, BTBT primarily occurs at gate 1 and gate 2. This method enables the identification of the relative area contributing to the effect. The results confirm that ML models can effectively discern causative factors, facilitating optimization by establishing associ ations with each parameter.

Fig. 4. (a) R2value of each gate areas; (b) Electron BTBT generation of channel and separated gate area.
../../Resources/ieie/JSTS.2024.24.2.96/fig4.png

IV. MULTIPLE CAUSES OF VARIATION

To validate the previously developed ML model, it is crucial to examine whether it accounts for various effects. In contrast to MOSFETs, TFETs have been shown to receive inversion carriers from the drain. This inversion charge inhibits channel band bending, a phenomenon influenced by randomly distributed metal grains in the gate [26]. Additionally, when the Vd is low, several additional phenomena come into play. These include ambipolar characteristics, directly influenced by BTBT, and super-linear onset, observable by examining ambipolar current (IAMB), Vth, and Von [27].

The ML model for this investigation derived input from the structure and energy diagram, similarly shown as in Section 2. However, the output parameters were derived from Fig. 5(a) (Id-Vg curve when Vd = 0.1 V). The selected output parameters for the ML model were IAMB, Vth, and Von, with IAMB measured when Vg = -1 V. The built model exhibited a high R2 value for predictions at each parameter, as shown in Fig. 5(b)-(d). The R2 values for each parameter— IAMB (0.9901), Vth (0.9901), and Von (0.9927)—demonstrate that the ML model can accurately predict various phenomena with a high level of confidence.

Fig. 5. (a)Id-Vgcurve of nanowire TFET atVd= 0.1 V; (b) R2value ofIAMB; (c) R2value ofVth; (d) R2value ofVon; (e) Number of 4.4-eV grains; (f) R2value of prediction ofVonand testVonof new 10 exceptional data.
../../Resources/ieie/JSTS.2024.24.2.96/fig5.png

V. PREDICTION OF EXCEPTIONAL SITUATION

While most device structures exhibit average performance, devices such as CMOS or static random access memory (SRAM) can face performance challenges in exceptional situations [28]. This paper leverages ML to predict and address such exceptional scenarios that deviate significantly from the norm. ML models, developed in Section 2 and Section 5, were trained using data from Fig. 5(e).

Each gate in the ML model comprises grains with a 40% probability, featuring 4.4-eV grains whereas typical structures consist of 12 to 13 (average of 12.678) grains. To simulate an exceptional situation, 10 new structures were created, each with a 75% probability of acquiring 4.4-eV grains. In the existing structures, several 4.4-eV grains were located at positions 26, 22, 24, 20, 27, 24, 24, 26, 20, and 26, typically situated in the red part of Fig. 5(e). These structures were not part of the ML model training.

To make predictions, the grains and energy diagram of the structures were normalized using MinMaxScaler, a process consistent with the existing ML model inputs. The prediction of the input and output of the ML model was also normalized. Comparing the predicted Von with the actual results, a high R2 value of 0.9927 was achieved [Fig. 5(f)]. This outcome demonstrates that the ML model can effectively predict various exceptional situations.

VI. CONCLUSIONS

This study proposes the optimization of the ML model to establish a correlation between WFV and the variation of Von in TFETs. Through the optimization and analysis of the ML model, it becomes evident that TFET performance is more affected by the number of gate regions than by the entire grain structure of the gate. This implies that only a few grains exhibit a high correlation with TFET's Von. The ML model, relying on diverse data and exhibiting a strong dependence on parameters like Vd, demonstrates the potential for identifying multiple influencing factors. Furthermore, the ML model predicts that a gate with a higher proportion of 4.4 eV WF is likely to provide insights into the underlying causes. Ultimately, the results indicate the potential for prediction and analysis in the semiconductor process or simulation, particularly with sufficient and diverse data.

ACKNOWLEDGMENTS

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No.2022R1A2C1093201). The EDA tool was supported by the IC Design Education Center (IDEC), KOREA

References

1 
C.-T. Sah, “Characteristics of the metal-oxide-semiconductor transistors,” IEEE Trans. Electron Devices, Vol. 11, No. 7, pp. 324-345, 1964.DOI
2 
Q. Zhang, W. Zhao, and A. Seabaugh, “Low-subthreshold-swing tunnel transistors,” IEEE Electron Device Lett., Vol. 27, No. 4, pp. 297-300, 2006.DOI
3 
J. H. Kim, S. Kim, and B.-G. Park, “Double-gate TFET with vertical channel sandwiched by lightly doped Si,” IEEE Trans. Electron Devices, Vol. 66, No. 4, pp. 1656-1661, 2019.DOI
4 
A. C. Seabaugh and Q. Zhang, “Low-voltage tunnel transistors for beyond CMOS logic,” Proc. IEEE, Vol. 98, No. 12, pp. 2095-2110, 2010.DOI
5 
R. Gandhi, Z. Chen, N. Singh, K. Banerjee, and S. Lee, “Vertical Si-Nanowire $ n $-Type Tunneling FETs With Low Subthreshold Swing (≤ 50 [mV/decade] at Room Temperature,” IEEE Electron Device Lett., Vol. 32, No. 4, pp. 437-439, 2011.DOI
6 
W. Y. Choi, B.-G. Park, J. D. Lee, and T.-J. K. Liu, “Tunneling field-effect transistors (TFETs) with subthreshold swing (SS) less than 60 mV/dec,” IEEE Electron Device Lett., Vol. 28, No. 8, pp. 743-745, 2007.DOI
7 
N. Damrongplasit, C. Shin, S. H. Kim, R. A. Vega, and T.-J. K. Liu, “Study of random dopant fluctuation effects in germanium-source tunnel FETs,” IEEE Trans. Electron Devices, Vol. 58, No. 10, pp. 3541-3548, 2011.DOI
8 
A. Asenov, S. Kaya, and A. R. Brown, “Intrinsic parameter fluctuations in decananometer MOSFETs introduced by gate line edge roughness,” IEEE Trans. Electron Devices, Vol. 50, No. 5, pp. 1254-1260, 2003.DOI
9 
J. H. Kim, H. W. Kim, Y. S. Song, S. Kim, and G. Kim, “Analysis of current variation with work function variation in l-shaped tunnel-field effect transistor,” Micromachines, Vol. 11, No. 8, p. 780, 2020.DOI
10 
K. M. Choi, S. K. Kim, and W. Y. Choi, “Influence of number fluctuation and position variation of channel dopants and gate metal grains on tunneling field-effect transistors (TFETs),” J. Nanosci. Nanotechnol., Vol. 16, No. 5, pp. 5255-5258, 2016.DOI
11 
Y.-C. Yeo, T.-J. King, and C. Hu, “Metal-dielectric band alignment and its implications for metal gate complementary metal-oxide-semiconductor tech-nology,” J. Appl. Phys., Vol. 92, No. 12, pp. 7266-7271, 2002.DOI
12 
H. Nam and C. Shin, “Study of high-k/metal-gate work-function variation using Rayleigh distribution,” IEEE Electron Device Lett., Vol. 34, No. 4, pp. 532-534, 2013.DOI
13 
G. Kim, J. H. Kim, J. Kim, and S. Kim, “Analysis of work-function variation effects in a tunnel field-effect transistor depending on the device structure,” Appl. Sci., Vol. 10, No. 15, p. 5378, 2020.DOI
14 
C. C. Hobbs et al., “Fermi-level pinning at the polysilicon/metal-oxide interface-Part II,” IEEE Trans. Electron Devices, Vol. 51, No. 6, pp. 978-984, 2004.DOI
15 
A. Ravindran, A. George, C. S. Praveen, and N. Kuruvilla, “Gate all around nanowire TFET with high ON/OFF current ratio,” Mater. Today Proc., Vol. 4, No. 9, pp. 10637-10642, 2017.DOI
16 
Y. Lee, H. Nam, J.-D. Park, and C. Shin, “Study of work-function variation for high- κ metal-gate Ge-Source tunnel field-effect transistors,” IEEE Trans. Electron Devices, vol. 62, no. 7, pp. 2143-2147, 2015.DOI
17 
H. F. Dadgour, K. Endo, V. K. De, and K. Banerjee, “Grain-orientation induced work function variation in nanoscale metal-gate transistors—Part II: Implications for process, device, and circuit design,” IEEE Trans. Electron Devices, Vol. 57, No. 10, pp. 2515-2525, 2010.DOI
18 
T. M. Mitchell and T. M. Mitchell, Machine learning, vol. 1, no. 9. McGraw-hill New York, 1997.URL
19 
J. Lim and C. Shin, “Machine learning (ML)-based model to characterize the line edge roughness (LER)-induced random variation in FinFET,” IEEE Access, Vol. 8, pp. 158237-158242, 2020.DOI
20 
Y. S. Bankapalli and H. Y. Wong, “TCAD augmented machine learning for semiconductor device failure troubleshooting and reverse engineering,” in 2019 International Conference on Simulation of Semiconductor Processes and Devices (SISPAD), 2019, pp. 1-4.DOI
21 
Y.-C. Wu and Y.-R. Jhan, “Introduction of synopsys sentaurus TCAD simulation,” in 3D TCAD Simulation for CMOS Nanoeletronic Devices, Springer, 2018, pp. 1-17.DOI
22 
E. P. Gyftopoulos and J. D. Levine, “Work function variation of metals coated by metallic films,” J. Appl. Phys., Vol. 33, No. 1, pp. 67-73, 1962.DOI
23 
G.-B. Huang, “Learning capability and storage capacity of two-hidden-layer feedforward networks,” IEEE Trans. neural networks, Vol. 14, No. 2, pp. 274-281, 2003.DOI
24 
Z. Zhang, “Improved adam optimizer for deep neural networks,” in 2018 IEEE/ACM 26th International Symposium on Quality of Service (IWQoS), 2018, pp. 1-2.DOI
25 
A. F. Agarap, “Deep learning using rectified linear units (relu),” arXiv Prepr. arXiv1803.08375, 2018.DOI
26 
H. W. Kim and J. H. Kim, “Study on the Influence of Drain Voltage on Work Function Variation Characteristics in Tunnel Field-effect Transistor,” J. Semicond. Technol. Sci., Vol. 20, No. 6, pp. 558-564, 2020.DOI
27 
L. De Michielis, L. Lattanzio, and A. M. Ionescu, “Understanding the superlinear onset of tunnel-FET output characteristic,” IEEE Electron Device Lett., Vol. 33, No. 11, pp. 1523-1525, 2012.DOI
28 
K. Bernstein, R. K. Cavin, W. Porod, A. Seabaugh, and J. Welser, “Device and architecture outlook for beyond CMOS switches,” Proc. IEEE, Vol. 98, No. 12, pp. 2169-2184, 2010.DOI
Tae Hyun Hwang
../../Resources/ieie/JSTS.2024.24.2.96/au1.png

Tae Hyun Hwang was born in Busan, Korea, on 1999. He received the B.S. degree in the Department of Electronic and Electrical Engineering from Pukyong National University, Korea, in 2024. Currently, he is working as a manager from Power Electronics R&D Team at LS Electric.

Sangwan Kim
../../Resources/ieie/JSTS.2024.24.2.96/au2.png

Sangwan Kim received the B.S., M.S., and the Ph.D. degrees in Electrical from Seoul National University, Seoul, Republic of Korea, in 2006, 2008, and 2014, respectively. He had been a post-doctoral scholar at the Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, USA, from 2014 to 2017. He had been with the Department of Electrical and Computer Engineering, Ajou University, Suwon, Republic of Korea, as Assistant/Associate Professor from 2017 to 2022. Since 2022, he has been a Faculty Member with Sogang University, Seoul, Republic of Korea, where he is currently an Associate Professor with the Department of Electronic Engineering. His current research interest includes ultra-low power logic devices, future memory devices, synaptic devices and their applications.

Garam Kim
../../Resources/ieie/JSTS.2024.24.2.96/au3.png

Garam Kim recived the B.S. and the Ph.D. degrees in electrical engineering from Seoul National University, Seoul, Korea, in 2008 and 2014, respectively. He worked as a senior engineering at Samsung Electronics from 2014 to 2019. In 2019, he joined the Department of Electronic Engineering at Myongji University, Yongin, where he is currently an assistant professor. His current research interests include GaN-based LEDs, tunnel FETs, neuromorphic devices, capacitor-less 1T DRAMs, and GAN HEMT.

Hyunwoo Kim
../../Resources/ieie/JSTS.2024.24.2.96/au4.png

Hyunwoo Kim received the B.S. degree from the Kyungpook National University (KNU), Daegu, South Korea, in 2008, and the M.S. and Ph.D. degrees in Electrical Engineering from Seoul National University (SNU), Seoul, in 2010 and 2015, respectively. He had worked at Samsung Electronics as a senior researcher from 2015 to 2021, Hwaseong, Korea. Since 2023, he has been a Faculty Member with Konkuk University (KU), Seoul, Korea, where he is currently an Assistant Professor with the Department of Electrical and Electronics Engineering. His current interests for research include Foundry Logic CMOS Devices, Ferroelectric Devices, and Low Power Applications.

Jang Hyun Kim
../../Resources/ieie/JSTS.2024.24.2.96/au5.png

Jang Hyun Kim completed his Bachelor’s degree in Electrical and Electronic Engineering at KAIST (Korea Advanced Institute of Science and Technology) from March 2005 to August 2009. He then pursued his Master’s degree in Dept. of Electrical and Computer Engineering at Seoul National University, from September 2009 to August 2011. Continuing his academic journey, he obtained his Doctorate degree in Dept. of Electrical and Computer Engineering. After completing his education, Jang Hyun Kim worked as a Development Researcher for DRAM (Dynamic Random-Access Memory) at SK hynix from September 2016 to February 2020. Subsequently, he served as an Assistant Professor in the Department of Electrical Engineering at Pukyong National University from March 2020 to February 2023. Currently, he holds the position of Assistant Professor in the Department of Electronic Engineering at Ajou University, starting from February 2023. His current research interests include logic semiconductor devices and power semiconductor devices.