Heebum Kang

Heebum Kang

Auto-generated profile, based on publicly available information

Heebum Kang

School of Electronic and Electrical Engineering, Kyungpook National University, Daegu, South Korea

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 August 2021 in Applied Physics Letters
Verified
Journal Article
Closed access
PDF file
HKHeebum Kang
Heebum Kang
JWJiyong Woo
Jiyong Woo
An analogous change in lateral channel current from source to drain in three-terminal synaptic devices actuated by mobile ions vertically provided from a gate can enhance neuromorphic computing performances. We demonstrate a gradually tuned channel current in a fully complementary metal–oxide–semiconductor compatible HfOx/WOx stack with Cu ions. By examining each layer in the three-terminal device, such as the channel, electrolyte, and mobile ion supplier, we identify which device structure can modulate the channel current effectively using mobile ions. Our findings reveal that the gate-tunable channel current response can be solely achieved when the Cu ions are not locally formed but migrate throughout the HfOx electrolyte. The linear dependence of the analog current operation on the channel width further proves the area-switching mechanism. The importance of ion movement can be indirectly verified from the uncontrollable channel currents using either Ag ions with faster mobility than Cu ions or a local path is created because of the thinned HfOx electrolyte.

11 April 2023 in IEEE Transactions on Electron Devices
Verified
Journal Article
Closed access
HKHyun Wook Kim
Hyun Wook Kim
SJSeonuk Jeon
Seonuk Jeon
HKHeebum Kang
Heebum Kang
EHEunryeong Hong
Eunryeong Hong
NKNayeon Kim
Nayeon Kim
JWJiyong Woo
Jiyong Woo
Oscillatory neural networks (ONNs) directly emulate signal communication between biological neurons in the human brain by encoding the data in phase domain, enabling energy-efficient associative memory. An oscillation neuron (ON) element that generates continuous voltage spikes with a specific frequency needs to be designed for hardware implementation. Thus, we systematically investigate the role of the ON in edge detection in ONN systems through simulation. First, a threshold switch is experimentally fabricated for the ON using niobium oxide (NbO $_{\textit{x}}\text{)}$ material, and voltage oscillation is realized in HSPICE and MATLAB. Subsequently, we examine how each voltage oscillation in a coupled-ON system, in which two NbO $_{\textit{x}}$ -based ONs are connected with a coupling resistance, is mutually synchronized. Simulation results reveal that a small (or large) coupling resistance strengthens the in-phase (or out-of-phase) synchronization of the two independent oscillations. The synchronized phase expressed in the form of period is found to be adjusted by tuning various components. As two clearly distinguishable phases are obtained, ONN systems, where multiple ONs are cross-coupled, can be utilized for edge detection during image processing. Patterns are trained using Hebbian learning rule in an ONN system comprising ten ONs, and a feature of the handwritten digit image is accurately extracted. Moreover, the feasibility of accelerating the edge detection step is further explored through various engineering approaches to change the characteristics of the NbO $_{\textit{x}}$ -based ONs.

21 March 2022 in Applied Physics Letters
Verified
Journal Article
Closed access
PDF file
HKHeebum Kang
Heebum Kang
HKHyun Wook Kim
Hyun Wook Kim
EHEun Ryeong Hong
Eun Ryeong Hong
JWJiyong Woo
Jiyong Woo
We demonstrate the synaptic characteristics of analogously modulated channel currents in Cu-ion-actuated electrochemical RAM (ECRAM) based on an HfOx electrolyte and a WOx channel. Uncontrolled synaptic response is found as a function of the gate pulse when a Cu-rich gate electrode delivers mobile ions, presumably due to many ions injected from the infinite ion reservoir. As a result, we propose a CuOx oxide electrode to limit ion sources, which is indirectly validated by a physical examination of the degree of chemical bonding between Cu and oxygen, thereby boosting gate controllability over the channel. In addition, the HfOx electrolyte needs to be designed to facilitate the adequate migration of Cu ions, considering thickness and film quality. Using material stack engineering, the channel current of optimized CuOx/HfOx/WOx ECRAM can be steadily tuned via repeated identical gate pulses. The channel current and its change are proportional to the device area and the amount of migrated ions relevant to the gate pulse conditions, respectively. The homogeneous flow of ions across the entire area can, thus, be used to explain the obtained analog switching. The gate-controllable synaptic behavior of the ECRAM accelerates deep neural network training based on backpropagation algorithms. An improved pattern recognition accuracy of ∼88% for handwritten digits is achieved by linearly tuned multiple current states with more than 100 pulses and asymmetric gate voltage conditions in a three-layer neural network validated in simulation.

25 April 2022 in IEEE Journal of the Electron Devices Society
Verified
Journal Article
Open access
HKHyun Wook Kim
Hyun Wook Kim
HKHeebum Kang
Heebum Kang
EHEunryeong Hong
Eunryeong Hong
JWJiyong Woo
Jiyong Woo
The reversible transition between the on and off states of threshold switches under a constant pulse generates voltage oscillation, which can be exploited for compact neuron element in neuromorphic systems. Because the transition voltages play an important role in the oscillation behavior, area and thickness scaling analysis of NbO x -based devices is performed to identify the underlying mechanism. The threshold voltage ( $V_{th}$ ) is sensitive to the device area, indicating that the on state of the device is achieved by the local formation of a conductive phase. On the other hand, the area-independent hold voltage ( $V_{\mathrm{ hold}}$ ) becomes smaller due to the higher compliance current and increased temperature ambient, which implies that spontaneous dissolution of the phase is retarded. Through HSPICE simulation, we reveal that the greater difference between $\text{V}_{\mathrm{ hold}}$ and $\text{V}_{\mathrm{ th}}$ enables a high degree of freedom of oscillation frequency modulation.

22 February 2023 in IEEE Transactions on Electron Devices
Verified
Journal Article
Closed access
SJSeonuk Jeon
Seonuk Jeon
EHEunryeong Hong
Eunryeong Hong
HKHeebum Kang
Heebum Kang
HKHyun Wook Kim
Hyun Wook Kim
NKNayeon Kim
Nayeon Kim
JWJiyong Woo
Jiyong Woo
As simple convolution computation is intensively and iteratively performed to extract features from input images, cross-point arrays with resistive random access memory (RRAM) serving as a kernel weight can accelerate the relevant mathematical operations in hardware. However, considering actual RRAM characteristics, either variability or unexpected permanent failure from the filamentary switching mechanism is observed, degrading recognition performance. This study investigates the impact of fault in a conventional kernel structure, where two adjacent columns in the array represent a single weight, on feature extraction using MATLAB. First, the fault types of HfO $_{\textit{x}}$ -based multilevel RRAM is categorized. The results reveal that the unidirectional fault of RRAM primarily worsens the accuracy of image recognition. This is because the subtraction of negative weights from positive ones is crucial for identifying the edges of images through convolution operations. Therefore, we exploit a kernel structure, in which a single column dedicated to negative weights is located next to a matrix of positive weights. In addition, we reduce the weight precision for negative weights, while quantizing positive weights to higher bits. By mitigating the subtraction errors achieved by the kernel structure with hybrid precision, we improved fault tolerance, minimizing accuracy degradation.

14 October 2024 in IEEE Electron Device Letters
Verified
Journal Article
Closed access
HKHeebum Kang
Heebum Kang
KLKyumin Lee
Kyumin Lee
SHSeungkwon Hwang
Seungkwon Hwang
HHHyunsang Hwang
Hyunsang Hwang
We introduce a novel all-solid-state Mg-ion-based electrochemical RAM (Mg-ECRAM) that utilizes a highly stable MgF2 electrolyte known for its high ionic conductivity (σ ion ) and low electrical conductivity (σ). Additionally, crystalline WO 2.8 (C-WO 2.8 ) is used as the channel material because of its excellent ion diffusivity (D ion ). Comprehensively, our findings reveal nearly perfect weight update linearity and exceptional retention capabilities, lasting approximately six years. These results indicate that Mg-ions are suitable for ECRAM systems, offering desirable and dependable synaptic properties. Moreover, the physical intercalation of Mg-ions into the WO 2.8 channel is confirmed in real-time by the sequential modulation of Raman peaks, which correspond to the levels of potentiation or depression.