Article
Physics, Applied
Subhajit Ghosh, Dinusha Herath Mudiyanselage, Sergey Rumyantsev, Yuji Zhao, Houqiang Fu, Stephen Goodnick, Robert Nemanich, Alexander A. Balandin
Summary: In this study, we investigated the low-frequency electronic noise in beta-(AlxGa1-x)(2)O-3 Schottky barrier diodes. The noise spectrum exhibited 1/f behavior with Lorentzian bulges at intermediate current levels. The normalized noise spectral density was determined to be around 10(-12) cm(2)/Hz at a current density of 1 A/cm(2) and a frequency of 10 Hz. We observed random telegraph signal noise at intermediate currents, which was attributed to defects near the Schottky barrier.
APPLIED PHYSICS LETTERS
(2023)
Article
Physics, Applied
Ya-Yi Chen, Yuan Liu, Yuan Ren, Zhao-Hui Wu, Li Wang, Bin Li, Yun-Fei En, Yi-Qiang Chen
Summary: This paper investigates the forward bias conduction and LFN characteristics of GaN SBDs, revealing the influences of Schottky barrier inhomogeneities and 1/f noise on noise levels. The temperature dependence of ideality factor and zero bias Schottky barrier height are analyzed, and flicker noise is identified as the main component of LFN in GaN SBDs.
MODERN PHYSICS LETTERS B
(2021)
Article
Engineering, Electrical & Electronic
Wonjun Shin, Dongseok Kwon, Jong-Ho Bae, Suhwan Lim, Byung-Gook Park, Jong-Ho Lee
Summary: The study reveals that the 1/f noise origin of GSD varies depending on the operation mode and current region, leading to different effects of P/E cycling. Specifically, the n-type mode GSD operating in the high I-O region shows the largest noise increase after 10⁴ P/E cycling.
IEEE ELECTRON DEVICE LETTERS
(2021)
Review
Chemistry, Physical
Shi-Jun Liang, Yixiang Li, Bin Cheng, Feng Miao
Summary: In recent decades, advances in material science have allowed for precise control of material structures at the atomic scale and the realization of various low-dimensional heterostructures. These heterostructures have been widely used in electronic, spintronic, and optoelectronic devices with great success. Manipulating the dimensionality and physical properties of low-dimensional materials enables the design of a wide range of low-dimensional artificial heterostructures. The unique physical properties exhibited by these emerging low-dimensional heterostructures provide unprecedented opportunities for neuromorphic computing applications, laying the groundwork for future intelligent systems.
Article
Chemistry, Multidisciplinary
Shashank Misra, Leslie C. Bland, Suma G. Cardwell, Jean Anne C. Incorvia, Conrad D. James, Andrew D. Kent, Catherine D. Schuman, J. Darby Smith, James B. Aimone
Summary: The brain has provided powerful inspiration for computing architectures, and neuromorphic systems have the potential to improve scientific computing and artificial intelligence. The brain's stochasticity can be a source of inspiration for expanding neuromorphic computing to probabilistic applications. Current efforts in probabilistic computing focus on specific scales of microelectronics, but a co-design vision is proposed to operate devices in a stochastic regime and incorporate them into scalable neuromorphic architectures.
ADVANCED MATERIALS
(2023)
Article
Automation & Control Systems
Wonjun Shin, Kyung Kyu Min, Jong-Ho Bae, Jaehyeon Kim, Ryun-Han Koo, Dongseok Kwon, Jae-Joon Kim, Daewoong Kwon, Jong-Ho Lee
Summary: In recent years, the development of neuromorphic computing has faced the limitations of von Neumann architecture. Therefore, there is a growing demand for high-performance synaptic devices that possess high switching speeds, low power consumption, and multilevel conductance. Among various synaptic devices, ferroelectric tunnel junctions (FTJs) have emerged as promising candidates. While previous studies have focused on improving the reliability of FTJs to enhance synaptic behavior, the low-frequency noise (LFN) of FTJs and its impact on the learning accuracy in neuromorphic computing have not been thoroughly investigated. This study explores the LFN characteristics of FTJs fabricated on n- and p-type Si and evaluates the impact of 1/f noise on the learning accuracy of convolutional neural networks (CNNs). The results demonstrate that FTJs on p-type Si exhibit significantly lower 1/f noise than those on n-type Si. Consequently, the FTJs on p-type Si achieve a significantly higher learning accuracy (86.26%) compared to those on n-type Si (78.70%) due to their low-noise properties. This study provides valuable insights into the LFN characteristics of FTJs and offers a potential solution to enhance the performance of synaptic devices by drastically reducing 1/f noise.
ADVANCED INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Information Systems
Humberto Pereira Da Paz, Vinicius Santana Da Silva, Renan Diniz, Renan Trevisoli, Carlos Eduardo Capovilla, Ivan Roberto Santana Casella
Summary: This article focuses on improving low-power Radio Frequency Energy Harvesting (RFEH) designs using Schottky Barrier Diode (SBD) rectifiers. The study investigates the relationship between RF-DC Power Conversion Efficiency (PCE) and circuit temperature, specifically the non-linear behavior of the metal-semiconductor junction of SBDs. Through evaluation and modeling, two series RF rectifiers based on specific diodes are designed for efficient operation over different temperature ranges. The results demonstrate the stability of PCE for one of the diodes, despite overall limitations due to matching network losses.
Article
Materials Science, Multidisciplinary
Peng Shi, Dong Wang, Tongliang Yu, Ruofei Xing, Zhenfa Wu, Shishen Yan, Lin Wei, Yanxue Chen, Huixue Ren, Chunfeng Yu, Fangjun Li
Summary: A novel three-terminal synaptic transistor was designed and constructed, achieving nonvolatile conductance modulation through electrolyte gating and successfully mimicking synaptic learning functions.
MATERIALS & DESIGN
(2021)
Article
Chemistry, Multidisciplinary
Bin-Wei Yao, Jiaqiang Li, Xu-Dong Chen, Mei-Xi Yu, Zhi-Cheng Zhang, Yuan Li, Tong-Bu Lu, Jin Zhang
Summary: Emerging electrolyte-gated transistors (EGTs) with ion-storage layer, such as graphdiyne (GDY)/MoS2-based EGT, demonstrate robust stability, low energy consumption, and non-volatile characteristics, making them potential candidates for low-power neuromorphic computing beyond von Neumann architecture. These devices show promising analog switching performance, quasi-linear conductance update, and logic-in-memory functions, highlighting their potential for next-generation electronics with near-ideal accuracies.
ADVANCED FUNCTIONAL MATERIALS
(2021)
Review
Chemistry, Multidisciplinary
Shilei Dai, Xu Liu, Youdi Liu, Yutong Xu, Junyao Zhang, Yue Wu, Ping Cheng, Lize Xiong, Jia Huang
Summary: Living organisms possess a mysterious and powerful sensory computing system based on ion activity. The development of iontronic devices in recent years has provided a promising platform for simulating the sensing and computing functions of living organisms. These devices can generate, store, and transmit signals by adjusting ion concentration and distribution, bridging biology and electronics, and offering advantages in sensing and recognition. This review provides an overview of neuromorphic sensory computing by iontronic devices, highlighting concepts and breakthroughs, and discusses challenges and future directions.
ADVANCED MATERIALS
(2023)
Article
Engineering, Electrical & Electronic
Xiaowei Feng, Surya Abhishek Singaraju, Hongrong Hu, Gabriel Cadilha Marques, Tongtong Fu, Peter Baumgartner, Daniel Secker, Mehdi B. Tahoori, Jasmin Aghassi-Hagmann
Summary: This study characterized the low-frequency noise of inkjet-printed electrolyte-gated thin-film transistors and identified the dominating noise generation mechanism. Benchmark analysis on the noise level of various thin-film technologies showed that the electrolyte-gating approach effectively reduces transistor noise levels.
IEEE ELECTRON DEVICE LETTERS
(2021)
Article
Nanoscience & Nanotechnology
Fengben Xi, Andreas Grenmyr, Jiayuan Zhang, Yi Han, Jin Hee Bae, Detlev Grutzmacher, Qing-Tai Zhao
Summary: Neuromorphic computing employs artificial synapses to transfer information between neurons. Conventional artificial synapses with homosynaptic plasticity face positive feedback loop problem, requiring synapses with heterosynaptic plasticity. This study presents complementary metal-oxide-semiconductor compatible artificial synapses based on FEMOD on silicon, allowing heterosynaptic plasticity with multi-functionalities, high endurance, low power consumption, and high speed. The proposed device structure performs multi-functions of biological synapse and Boolean logic, providing high potential for future large scale and low power neuromorphic computing applications.
ADVANCED ELECTRONIC MATERIALS
(2023)
Article
Chemistry, Multidisciplinary
Xiu Fang Lu, Yishu Zhang, Naizhou Wang, Sheng Luo, Kunling Peng, Lin Wang, Hao Chen, Weibo Gao, Xian Hui Chen, Yang Bao, Gengchiau Liang, Kian Ping Loh
Summary: The study demonstrates a filament-based memristor using p-type SnS as the resistive switching material, exhibiting superlative metrics such as low power consumption, fast switching speed, high endurance switching cycles, and a large on/off ratio. Chip-level simulations reveal an on-chip learning accuracy of 87.76% for image classifications, attributed to the ultrafast and low energy switching of p-type SnS compared to n-type transition metal dichalcogenides.
Article
Nanoscience & Nanotechnology
Chang Liu, Yanghao Wang, Teng Zhang, Rui Yuan, Yuchao Yang
Summary: Neuromorphic computing aims to connect cognitive behaviors with efficient computing systems in a biologically inspired way. Attention mechanism, an important cognitive behavior for filtering and regulating spatio-temporal information, can be efficiently processed using the dynamic capabilities of emerging neuromorphic devices. A basic top-down attention computing component comprising a synaptic transistor and a neuron is proposed and demonstrated to effectively filter and control information. The component exhibits new dynamic circuit behaviors, such as conductance oscillation and activate function switching, offering a power and area-saving method for constructing complex neuromorphic systems.
ADVANCED ELECTRONIC MATERIALS
(2023)
Review
Physics, Applied
Jun-Seok Ro, Hye-Min An, Hea-Lim Park
Summary: The limitations of von Neumann computing systems have been addressed by utilizing neuromorphic devices, particularly electrolyte-gated synaptic transistors (EGSTs), which operate through ion movement in electrolytes. EGSTs are desirable for neuromorphic computing due to their efficient energy consumption and biocompatibility. Recent studies on EGSTs can be classified into four main categories: synaptic plasticity, fast switching speed, low energy consumption, and biocompatibility. To extend the applications of EGSTs, additional requirements and limitations need to be addressed.
JAPANESE JOURNAL OF APPLIED PHYSICS
(2023)
Article
Engineering, Electrical & Electronic
Sung Yun Woo, Dongseok Kwon, Byung-Gook Park, Jong-Ho Lee, Jong-Ho Bae
Summary: A positive-feedback (PF) device and its operation scheme for implementing pulse width modulation (PWM) function were demonstrated. By adjusting the charge stored in the n(-) floating body (Q(n)), the potential of the floating body changes linearly with time. The voltage-to-pulse width conversion for PWM function was achieved by utilizing the linear time-varying property of Q(n) and the gate bias dependency of Q(th). This approach allows for the reduction of the area required for a PWM neuron.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Ryun-Han Koo, Wonjun Shin, Kyung Kyu Min, Dongseok Kwon, Dae Hwan Kim, Jae-Joon Kim, Daewoong Kwon, Jong-Ho Lee
Summary: We investigate the effects of temperature and the number of cycles on remnant polarization and carrier transport process to determine the factors that determine the tunneling electroresistance (TER) of the ferroelectric tunnel junction (FTJ). Our fabricated FTJs have a metal/ferroelectric/insulator/semiconductor structure. It is found that the remnant polarization increases with increasing temperature and number of cycles due to oxygen vacancy redistribution. However, the increased remnant polarization does not improve the TER ratio at higher temperature and number of cycles. Using current-voltage characterization and low-frequency noise spectroscopy, we reveal that the carrier transport process at the interface between the ferroelectric and dielectric layers plays a more important role in determining the TER ratio than remnant polarization.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Wonjun Shin, Ryun-Han Koo, Kyung Kyu Min, Dongseok Kwon, Jae-Joon Kim, Daewoong Kwon, Jong-Ho Lee
Summary: We demonstrate that the resistance switching of an undoped hafnium oxide-based ferroelectric tunnel junction is influenced by both ferroelectric domain switching and the redistribution of oxygen vacancies within the hafnium oxide. The resistance switching mechanism varies depending on the program bias applied to the tunnel junction. Through low-frequency noise spectroscopy, we present a precise method for distinguishing two distinct resistance switching processes intrinsic to the tunnel junction.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
In-Seok Lee, Hyeongsu Kim, Min-Kyu Park, Joon Hwang, Ryun-Han Koo, Jae-Joon Kim, Jong-Ho Lee
Summary: This study proposes a novel XNOR-AND hybrid binary neural network (BNN) using a TFT-type synaptic device to reduce the area and power consumption of the synaptic array. Replacing some parts of the network from the XNOR operation to the AND operation allows for expressing weight and input with a single cell and single word line. When replacing the operation of the fully-connected (FC) layer with the AND operation, the accuracy of VGG9 BNN for CIFAR-10 datasets drops by approximately 1%, while the number of cells in the synaptic array decreases by 33.7%. Utilizing the previously proposed TFT-type synaptic devices, the proposed method reduces power consumption by around 25%.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Min-Kyu Park, Won-Mook Kang, Ryun-Han Koo, Jeong-Hyun Kim, Joon Hwang, Jong-Ho Bae, Jae-Joon Kim, Jong-Ho Lee
Summary: This study proposes and verifies a novel method of integrating an AND-type flash synaptic array with CMOS circuits. By reducing the number of masks and fabrication steps, the proposed method successfully integrates the synaptic array and CMOS peripheral circuits on a single wafer. This research is of great significance for the efficient implementation of hardware-based neural networks.
IEEE TRANSACTIONS ON ELECTRON DEVICES
(2023)
Article
Chemistry, Physical
Juyeong Pyo, Jong-Ho Bae, Sungjun Kim, Seongjae Cho
Summary: A three-terminal synaptic transistor based on IGZO was fabricated, and its chemical compositions and thicknesses were verified using transmission electron microscopy and energy dispersive spectroscopy. The synaptic devices demonstrated short-term memory behaviors such as excitatory post-synaptic current (EPSC), paired-pulse facilitation (PPF), short-term potentiation (STP), and short-term depression (STD). The IGZO-based three-terminal synaptic transistor could be controlled by the amplitude, width, and interval time of the pulses for implementing neuromorphic systems.
Article
Chemistry, Multidisciplinary
Wonjun Shin, Jaehyeon Kim, Gyuweon Jung, Suyeon Ju, Sung-Ho Park, Yujeong Jeong, Seongbin Hong, Ryun-Han Koo, Yeongheon Yang, Jae-Joon Kim, Seungwu Han, Jong-Ho Lee
Summary: Concerns about air quality, industrial gas leaks, and medical diagnostics are driving the demand for high-performance gas sensors. This study presents a novel method, low-frequency noise (LFN) spectroscopy, to achieve selective detection using a single FET-type gas sensor. The proposed system provides a new and efficient method capable of selectively detecting a target gas using in-memory-computed LFN spectroscopy, paving the way for further development in gas sensing systems.
Article
Physics, Applied
Dongseok Kwon, Hyeongsu Kim, Kyu-Ho Lee, Joon Hwang, Wonjun Shin, Jong-Ho Bae, Sung Yun Woo, Jong-Ho Lee
Summary: This work proposes positive feedback (PF) device-based synaptic devices for reliable binary neural networks (BNNs). The fabricated PF device shows a high on/off current ratio (2.69 x 10(7)) due to PF operation. The PF device has a charge-trap layer for adjusting the turn-on voltage (V-on) through program/erase operations and implementing long-term memory function. The steep switching characteristics of the PF device provide tolerance to retention time and turn-on voltage variation, enabling high accuracy (88.44% for CIFAR-10 image classification) in hardware-based BNNs using PF devices as synapses.
APPLIED PHYSICS LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Wonjun Shin, Sangwoo Kim, Ryun-Han Koo, Dongseok Kwon, Jae-Joon Kim, Deok-Hwang Kwon, Daewoong Kwon, Jong-Ho Lee
Summary: In this study, the low-frequency noise (LFN) characteristics of hafnium-zirconium ferroelectric junctionless poly-Si thin-film transistors (FE JL TFTs) with different channel lengths (Ls) were investigated. The results showed that the magnitude of 1/f noise decreased with a decrease in channel length, which was opposite to the trend observed in conventional FETs. Furthermore, it was observed that the protrusion of poly-Si was more vulnerable to damage from tensile stress in devices with longer channel lengths.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Ryun-Han Koo, Wonjun Shin, Kyung Kyu Min, Dongseok Kwon, Jae-Joon Kim, Daewoong Kwon, Jong-Ho Lee
Summary: We investigate the effect of post-metal annealing temperature (T-PMA) on ferroelectric (FE) resistive switching (RS) and non-FE RS in HfOx ferroelectric tunnel junctions. Through conductance analysis and low frequency noise spectroscopy, the effects of T-PMA on RS mechanisms are demonstrated. It is revealed that the non FE RS, redistribution of oxygen vacancies, is suppressed with an increase in T-PMA. The effects of different RS mechanisms on the tunneling electroresistance and cycling endurance characteristics are systematically investigated.
IEEE ELECTRON DEVICE LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Dongseok Kwon, Min-Kyu Park, Won-Mook Kang, Joon Hwang, Ryun-Han Koo, Jong-Ho Bae, Jong-Ho Lee
Summary: This work designs hardware-based ternary neural networks (TNNs) using TFT-type synaptic devices and analyzes the impact of leakage currents on the inference accuracy of TNNs. Based on the analysis, systematic optimization of the operating conditions is conducted to improve accuracy.
IEEE TRANSACTIONS ON ELECTRON DEVICES
(2023)
Article
Multidisciplinary Sciences
Dongseok Kwon, Sung Yun Woo, Kyu-Ho Lee, Joon Hwang, Hyeongsu Kim, Sung-Ho Park, Wonjun Shin, Jong-Ho Bae, Jae-Joon Kim, Jong-Ho Lee
Summary: In this work, a reconfigurable neuromorphic computing (NC) block using a flash-type synapse array, emerging positive feedback (PF) neuron devices, and CMOS peripheral circuits is proposed and experimentally demonstrated. The flash memory enables easy calibration of the NC block for output signals, and the super-steep switching characteristics of the PF neuron device reduce the area overhead of the NC block. The NC block shows high energy efficiency (37.9 TOPS/W) and high accuracy (91.80%) for CIFAR-10 image classification, outperforming previous works. This work showcases the engineering potential of integrating synapses and neurons in terms of system efficiency and high performance.
Article
Computer Science, Information Systems
Dongyeon Kang, Wonjung Kim, Jun Tae Jang, Changwook Kim, Jung Nam Kim, Sung-Jin Choi, Jong-Ho Bae, Dong Myong Kim, Yoon Kim, Dae Hwan Kim
Summary: In this paper, a synaptic transistor with a floating-gate (FG) node and an amorphous InGaZnO (IGZO) channel layer is proposed. The device emulates the neuroplasticity functions of both short-term memory (STM) and long-term memory (LTM) through the control of the amplitude and the number of input pulses. The STM occurs when the input amplitude is relatively small (< 9 V) by ion movement in the gate dielectrics, and the LTM occurs when the input amplitude is relatively large (> 10 V) by storing electrons in the FG. Increasing the number of input pulses allows for longer information storage. The FG IGZO synaptic transistor could be a promising device solution for brain-inspired computing systems.
Article
Automation & Control Systems
Wonjun Shin, Kyung Kyu Min, Jong-Ho Bae, Jaehyeon Kim, Ryun-Han Koo, Dongseok Kwon, Jae-Joon Kim, Daewoong Kwon, Jong-Ho Lee
Summary: In recent years, the development of neuromorphic computing has faced the limitations of von Neumann architecture. Therefore, there is a growing demand for high-performance synaptic devices that possess high switching speeds, low power consumption, and multilevel conductance. Among various synaptic devices, ferroelectric tunnel junctions (FTJs) have emerged as promising candidates. While previous studies have focused on improving the reliability of FTJs to enhance synaptic behavior, the low-frequency noise (LFN) of FTJs and its impact on the learning accuracy in neuromorphic computing have not been thoroughly investigated. This study explores the LFN characteristics of FTJs fabricated on n- and p-type Si and evaluates the impact of 1/f noise on the learning accuracy of convolutional neural networks (CNNs). The results demonstrate that FTJs on p-type Si exhibit significantly lower 1/f noise than those on n-type Si. Consequently, the FTJs on p-type Si achieve a significantly higher learning accuracy (86.26%) compared to those on n-type Si (78.70%) due to their low-noise properties. This study provides valuable insights into the LFN characteristics of FTJs and offers a potential solution to enhance the performance of synaptic devices by drastically reducing 1/f noise.
ADVANCED INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Dongseok Kwon, Sung Yun Woo, Joon Hwang, Hyeongsu Kim, Jong-Ho Bae, Wonjun Shin, Byung-Gook Park, Jong-Ho Lee
Summary: Neuromorphic hardware using nonvolatile analog synaptic devices can reduce energy and time consumption for large-scale vector-matrix multiplication operations. However, existing training methods have reduced accuracy and high training costs. This study proposes a novel hybrid training method that efficiently trains the hardware using nonvolatile analog memory cells, demonstrating high performance in the fabricated hardware.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)