Linear Conductance Modulation in Aluminum Doped Resistive Switching Memories for Neuromorphic Computing

IF 2.1 4区材料科学 Q3 MATERIALS SCIENCE, MULTIDISCIPLINARY

Electronic Materials Letters Pub Date : 2024-08-09 DOI:10.1007/s13391-024-00516-w

Young-Woong Song, Junseo Lee, Sein Lee, Wooho Ham, Jeong Hyun Yoon, Jeong-Min Park, Taehoon Sung, Jang-Yeon Kwon

{"title":"Linear Conductance Modulation in Aluminum Doped Resistive Switching Memories for Neuromorphic Computing","authors":"Young-Woong Song, Junseo Lee, Sein Lee, Wooho Ham, Jeong Hyun Yoon, Jeong-Min Park, Taehoon Sung, Jang-Yeon Kwon","doi":"10.1007/s13391-024-00516-w","DOIUrl":null,"url":null,"abstract":"<div><p>With the advent of artificial intelligence (AI), automated machines could replace human labor in the near future. Nevertheless, AI implementation is currently confined to environments with huge power supplies and computing resources. Artificial neural networks are only implemented at the software level, which necessitates the continual retrieval of synaptic weights among devices. Physically constructing neural networks using emerging nonvolatile memories allows synaptic weights to be directly mapped, thereby enhancing the computational efficiency of AI. While resistive switching memory (RRAM) represents superior performances for in-memory computing, unresolved challenges persist regarding its nonideal properties. A significant challenge to the optimal performance of neural networks using RRAMs is the nonlinear conductance update. Ionic hopping of oxygen vacancy species should be thoroughly investigated and controlled for the successful implementation of RRAM-based AI acceleration. This study dopes tantalum oxide-based RRAM with aluminum, thus improving the nonlinear conductance modulation during the resistive switching process. As a result, the simulated classification accuracy of the trained network was significant improved.</p><h3>Graphical Abstract</h3>\n<div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":536,"journal":{"name":"Electronic Materials Letters","volume":"20 6","pages":"725 - 732"},"PeriodicalIF":2.1000,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Electronic Materials Letters","FirstCategoryId":"88","ListUrlMain":"https://link.springer.com/article/10.1007/s13391-024-00516-w","RegionNum":4,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}

引用次数: 0

Abstract

With the advent of artificial intelligence (AI), automated machines could replace human labor in the near future. Nevertheless, AI implementation is currently confined to environments with huge power supplies and computing resources. Artificial neural networks are only implemented at the software level, which necessitates the continual retrieval of synaptic weights among devices. Physically constructing neural networks using emerging nonvolatile memories allows synaptic weights to be directly mapped, thereby enhancing the computational efficiency of AI. While resistive switching memory (RRAM) represents superior performances for in-memory computing, unresolved challenges persist regarding its nonideal properties. A significant challenge to the optimal performance of neural networks using RRAMs is the nonlinear conductance update. Ionic hopping of oxygen vacancy species should be thoroughly investigated and controlled for the successful implementation of RRAM-based AI acceleration. This study dopes tantalum oxide-based RRAM with aluminum, thus improving the nonlinear conductance modulation during the resistive switching process. As a result, the simulated classification accuracy of the trained network was significant improved.

Graphical Abstract

Abstract Image

查看原文本刊更多论文

用于神经形态计算的掺铝电阻开关存储器中的线性电导调制

随着人工智能（AI）的出现，自动化机器在不久的将来可能会取代人类劳动。然而，人工智能的实现目前仅限于拥有巨大电力供应和计算资源的环境。人工神经网络只能在软件层面实现，这就需要在设备之间不断检索突触权重。利用新兴的非易失性存储器以物理方式构建神经网络，可以直接映射突触权重，从而提高人工智能的计算效率。虽然电阻开关存储器（RRAM）在内存计算方面性能优越，但它的非理想特性仍是未解决的难题。使用 RRAM 实现神经网络最佳性能的一个重大挑战是非线性电导更新。要成功实现基于 RRAM 的人工智能加速，就必须彻底研究和控制氧空位物种的离子跳跃。本研究在基于氧化钽的 RRAM 中掺入了铝，从而改善了电阻开关过程中的非线性电导调制。因此，训练网络的模拟分类准确性得到了显著提高。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Electronic Materials Letters 工程技术-材料科学：综合

CiteScore

4.70

自引率

20.80%

发文量

审稿时长

2.3 months

期刊介绍： Electronic Materials Letters is an official journal of the Korean Institute of Metals and Materials. It is a peer-reviewed international journal publishing print and online version. It covers all disciplines of research and technology in electronic materials. Emphasis is placed on science, engineering and applications of advanced materials, including electronic, magnetic, optical, organic, electrochemical, mechanical, and nanoscale materials. The aspects of synthesis and processing include thin films, nanostructures, self assembly, and bulk, all related to thermodynamics, kinetics and/or modeling.