• Journal of the Semiconductor & Display Technology
• /
• v.21 no.3
• /
• pp.50-56
• /
• 2022

In order to process a vast amount of data, there is demand for a new system with higher processing speed and lower energy consumption. To prevent 'memory wall' in von Neumann architecture, RRAM, which is a neuromorphic device, has been researched. In this paper, we summarize the features of RRAM and propose the device structure for characteristic improvement. RRAM operates as a synapse device using a change of resistance. In general, the resistance characteristics of RRAM are nonlinear and random. As synapse device, linearity and uniformity improvement of RRAM is important to improve learning recognition rate because high linearity and uniformity characteristics can achieve high recognition rate. There are many method, such as TEL, barrier layer, NC, high oxidation properties, to improve linearity and uniformity. We proposed a new device structure of TiN/Al doped TaO_x/AlO_x/Pt that will achieve high recognition rate. Also, with simulation, we prove that the improved properties show a high learning recognition rate.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.17 no.2
• /
• pp.174-179
• /
• 2017

We designed the CMOS analog integrate and fire (I&F) neuron circuit for driving memristor based on resistive-switching random access memory (RRAM). And we fabricated the RRAM device that have $HfO_2$ switching layer using atomic layer deposition (ALD). The RRAM device has gradual set and reset characteristics. By spice modeling of the synaptic device, we performed circuit simulation of synaptic device and CMOS neuron circuit. The neuron circuit consists of a current mirror for spatial integration, a capacitor for temporal integration, two inverters for pulse generation, a refractory part, and finally a feedback part for learning of the RRAM. We emulated the spike-timing-dependent-plasticity (STDP) characteristic that is performed automatically by pre-synaptic pulse and feedback signal of the neuron circuit. By STDP characteristics, the synaptic weight, conductance of the RRAM, is changed without additional control circuit.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2013.02a
• /
• pp.403-404
• /
• 2013

최근 scaling down의 한계에 부딪힌 DRAM과 Flash Memory를 대체하기 위한 차세대 메모리(Next Generation Memory)에 대한 연구가 활발히 진행되고 있다. ITRS (international technology roadmap for semiconductors)에 따르면 PRAM (phase change RAM), RRAM (resistive RAM), STT-MRAM (spin transfer torque magnetic RAM) 등이 차세대 메모리로써 부상하고 있다. 그 중 RRAM은 간단한 구조로 인한 고집적화, 빠른 program/erase 속도 (100~10 ns), 낮은 동작 전압 등의 장점을 갖고 있어 다른 차세대 메모리 중에서도 높은 평가를 받고 있다 [1]. 현재 RRAM은 주로 금속-산화물계(Metal-Oxide) 저항 변화 물질을 기반으로 연구가 활발하게 진행되고 있다. 하지만 근본적으로 공정 과정에서 산소에 의한 오염으로 인해 수율이 낮은 문제를 갖고 있으며, Endurance 및 Retention 등의 신뢰성이 떨어지는 단점이 있다. 따라서, 본 연구진은 산소 오염에 의한 신뢰성 문제를 근본적으로 해결할 수 있는 다양한 금속-질화물(Metal-Nitride) 기반의 저항 변화 물질을 제안해 연구를 진행하고 있으며, 우수한 열적 안정성($>450^{\circ}C$, 높은 종횡비, Cu 확산 방지 역할, 높은 공정 호환성 [2] 등의 장점을 가진 WN 박막을 저항 변화 물질로 사용하여 저항 변화 메모리를 구현하기 위한 연구를 진행하였다. WN 박막은 RF magnetron sputtering 방법을 사용하여 Ar/$N_2$ 가스를 20/30 sccm, 동작 압력 20 mTorr 조건에서 120 nm 의 두께로 증착하였고, E-beam Evaporation 방법을 통하여 Ti 상부 전극을 100 nm 증착하였다. I-V 실험결과, WN 기반의 RRAM은 양전압에서 SET 동작이 일어나며, 음전압에서 RESET 동작을 하는 bipolar 스위칭 특성을 보였으며, 읽기 전압 0.1 V에서 ~1 order의 저항비를 확보하였다. 신뢰성 분석 결과, $10^3$번의 Endurance 특성 및 $10^5$초의 긴 Retention time을 확보할 수 있었다. 또한, 고저항 상태에서는 Space-charge-limited Conduction, 저저항 상태에서는 Ohmic Conduction의 전도 특성을 보임에 따라 저항 변화 메카니즘이 filamentary conduction model로 확인되었다 [3]. 본 연구에서 개발한 WN 기반의 RRAM은 우수한 저항 변화 특성과 함께 높은 재료적 안정성, 그리고 기존 반도체 공정 호환성이 매우 높은 강점을 갖고 있어 핵심적인 차세대 메모리가 될 것으로 기대된다.

• Journal of the Microelectronics and Packaging Society
• /
• v.30 no.4
• /
• pp.32-43
• /
• 2023

Resistive Random Access Memory (RRAM), based on resistive switching characteristics, is emerging as a next-generation memory device capable of efficiently processing large amounts of data through its fast operation speed, simple device structure, and high-density implementation. Interface type resistive switching offer the advantage of low operation currents without the need for a forming process. Especially, for RRAM devices based on transition metal oxides, various studies are underway to enhance the memory characteristics, including precise material composition control and improving the reliability and stability of the device. In this paper, we introduce various methods, such as doping of heterogeneous elements, formation of multilayer films, chemical composition adjustment, and surface treatment to prevent degradation of interface type resistive switching properties and enhance the device characteristics. Through these approaches, we propose the feasibility of implementing high-efficient next-generation non-volatile memory devices based on improved resistive switching properties.

• Journal of IKEEE
• /
• v.26 no.4
• /
• pp.633-638
• /
• 2022

Rencently, neuromorphic systems of spiking neural networks (SNNs) that imitate the human brain have attracted attention. Neuromorphic technology has the advantage of high speed and low power consumption in cognitive applications and processing. Resistive random-access memory (RRAM) for SNNs are the most efficient structure for parallel calculation and perform the gradual switching operation of spike-timing-dependent plasticity (STDP). RRAM as synaptic device operation has low-power processing and expresses various memory states. However, the integration of RRAM device causes high switching voltage and current, resulting in high power consumption. To reduce the operation voltage of the RRAM, it is important to develop new materials of the switching layer and metal electrode. This study suggested a optimized new structure that is the Metal/Al₂O₃/HfO_x/SWCNTs/N+silicon (MOCS) with single-walled carbon nanotubes (SWCNTs), which have excellent electrical and mechanical properties in order to lower the switching voltage. Therefore, we show an improvement in the gradual switching behavior and low-power I/V curve of SWCNTs-based memristors.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2013.02a
• /
• pp.410-410
• /
• 2013

Next-generation nonvolatile memory (NVM) has attracted increasing attention about emerging NVMs such as ferroelectric random access memory, phase-change random access memory, magnetic random access memory and resistance random access memory (RRAM). Previous studies have demonstrated that RRAM is promising because of its excellent properties, including simple structure, high speed and high density integration. Many research groups have reported a lot of metal oxides as resistive materials like TiO2, NiO, SrTiO3 and ZnO [1]. Among them, the ZnO-based film is one of the most promising materials for RRAM because of its good switching characteristics, reliability and high transparency [2]. However, in many studies about ZnO-based RRAMs, there was a problem to get lower current level for reducing the operating power dissipation and improving the device reliability such an endurance and an retention time of memory devices. Thus in this paper, we investigated that highly reproducible bipolar resistive switching characteristics of W doped ZnO RRAM device and it showed low resistive switching current level and large ON/OFF ratio. This may be caused by the interdiffusion of the W atoms in the ZnO film, whch serves as dopants, and leakage current would rise resulting in the lowering of current level [3]. In this work, a ZnO film and W doped ZnO film were fabricated on a Si substrate using RF magnetron sputtering from ZnO and W targets at room temperature with Ar gas ambient, and compared their current levels. Compared with the conventional ZnO-based RRAM, the W doped ZnO ReRAM device shows the reduction of reset current from ~$10^{-6}$ A to ~$10^{-9}$ A and large ON/OFF ratio of ~$10^3$ along with self-rectifying characteristic as shown in Fig. 1. In addition, we observed good endurance of $10^3$ times and retention time of $10^4$ s in the W doped ZnO ReRAM device. With this advantageous characteristics, W doped ZnO thin film device is a promising candidates for CMOS compatible and high-density RRAM devices.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.14 no.3
• /
• pp.268-273
• /
• 2014

The forming, reset and set operation of bipolar resistive random access memory (RRAM) have been predicted by using a finite element (FE) model which includes interface effects. To the best of our knowledge, our bipolar RRAM model is applicable to realistic cell structure optimization because our model is based on the FE method (FEM) unlike precedent models.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.3
• /
• pp.663-669
• /
• 2014

The electrical characteristics of RRAM with different annealing temperature and thickness have been measured and discussed. The devices with Pt/Ti top electrode of 150nm, Pt bottom electrode of 150nm, $HfO_2$ oxide thickness of 45nm and 70nm have been fabricated. The fabricated device were classified by 3 different kinds according to the annealing temperature, such as non-annealed, annealed at $500^{\circ}C$ and annealed at $850^{\circ}C$. The set and reset voltages and the variation of resistance with temperatures have been measured as electrical properties. From the measurement, it was found that the set voltages were decreased and the reset voltage were increased slightly, and thus the sensing window was decreased with increasing of measurement temperatures. It was remarkable that the device annealed at $850^{\circ}C$ showed the best performances. Although the device with thickness of 45nm showed better performances in the point of the sensing window, the resistance of 45nm devices was large relatively in the low resistive state. It can be expected to enhance the device performances with ultra thin RRAM if the defect generation could be reduced at the $HfO_2$ deposition process.

• Journal of IKEEE
• /
• v.28 no.2
• /
• pp.234-238
• /
• 2024

This paper presents a study on the integration of 3D-stacked dual-tip RRAM with a CNN accelerator architecture, leveraging its low drive current characteristics and scalability in a 3D stacked configuration. The dual-tip structure is utilized in a parallel connection format in a synaptic array to implement multi-level capabilities. It is configured within a Network-on-chip style accelerator along with various hardware blocks such as DAC, ADC, buffers, registers, and shift & add circuits, and simulations were performed for the CNN accelerator. The quantization of synaptic weights and activation functions was assumed to be 16-bit. Simulation results of CNN operations through a parallel pipeline for this accelerator architecture achieved an operational efficiency of approximately 370 GOPs/W, with accuracy degradation due to quantization kept within 3%.

• Proceedings of the Materials Research Society of Korea Conference
• /
• 2007.11a
• /
• pp.63.2-63.2
• /
• 2007