Memristive Ion Dynamics to Enable Biorealistic Computing DOI Creative Commons
Ruoyu Zhao, Seung Ju Kim, Yichun Xu

et al.

Chemical Reviews, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 27, 2024

Conventional artificial intelligence (AI) systems are facing bottlenecks due to the fundamental mismatches between AI models, which rely on parallel, in-memory, and dynamic computation, traditional transistors, have been designed optimized for sequential logic operations. This calls development of novel computing units beyond transistors. Inspired by high efficiency adaptability biological neural networks, mimicking capabilities structures gaining more attention. Ion-based memristive devices (IMDs), owing intrinsic functional similarities their counterparts, hold significant promise implementing emerging neuromorphic learning algorithms. In this article, we review mechanisms IMDs based ion drift diffusion elucidate origins diverse dynamics. We then examine how these operate within different materials enable with various types switching behaviors, leading a wide range applications, from emulating components realizing specialized requirements. Furthermore, explore potential be modified tuned achieve customized dynamics, positions them as one most promising hardware candidates executing bioinspired algorithms unique specifications. Finally, identify challenges currently that hinder widespread usage highlight research directions could significantly benefit incorporating IMDs.

Language: Английский

2D materials-memristive devices nexus: From status quo to Impending applications DOI Creative Commons
Muhammad Muqeet Rehman, Yarjan Abdul Samad, Jahan Zeb Gul

et al.

Progress in Materials Science, Journal Year: 2025, Volume and Issue: unknown, P. 101471 - 101471

Published: Feb. 1, 2025

Language: Английский

Citations

2
Interface Element Accumulation‐Induced Single Ferroelectric Domain for High‐Performance Neuromorphic Synapse DOI Open Access
Xiaoqi Li, Jiaqi Liu, Fan Xu

et al.

Advanced Functional Materials, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 19, 2025

Abstract Ferroelectric (FE) synapses are promising for neuromorphic computing toward enhanced artificial intelligence systems. Nonetheless, there is a significant gap in understanding how to effectively tailor self‐polarization and its implications on synaptic device performance. Here, an approach using interfacial element accumulation reported the states of BaTiO 3 (BTO)/La 0.67 Sr 0.33 MnO (LSMO) FE heterostructure into single domain state. This configuration results demonstrated gradient distribution oxygen vacancies across film thickness, yielding extraordinary on/off ratio 10 7 Pt/BTO/LSMO diodes. giant resistive switching enables long‐term potentiation depression functions excellent linearity symmetry (with nonsymmetry factor as low 0.1), leading supervised learning ability associated neural network with high pattern recognition accuracy 95%. work provides simple design principle domain, which substantial enhancing performance computing.

Language: Английский

Citations

1
Physical mechanisms and integration design of memristors DOI

Mengna Wang,

Kun Wang, Bai Sun

et al.

Materials Today Nano, Journal Year: 2025, Volume and Issue: unknown, P. 100628 - 100628

Published: April 1, 2025

Language: Английский

Citations

0
Memristive Ion Dynamics to Enable Biorealistic Computing DOI Creative Commons
Ruoyu Zhao, Seung Ju Kim, Yichun Xu

et al.

Chemical Reviews, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 27, 2024

Conventional artificial intelligence (AI) systems are facing bottlenecks due to the fundamental mismatches between AI models, which rely on parallel, in-memory, and dynamic computation, traditional transistors, have been designed optimized for sequential logic operations. This calls development of novel computing units beyond transistors. Inspired by high efficiency adaptability biological neural networks, mimicking capabilities structures gaining more attention. Ion-based memristive devices (IMDs), owing intrinsic functional similarities their counterparts, hold significant promise implementing emerging neuromorphic learning algorithms. In this article, we review mechanisms IMDs based ion drift diffusion elucidate origins diverse dynamics. We then examine how these operate within different materials enable with various types switching behaviors, leading a wide range applications, from emulating components realizing specialized requirements. Furthermore, explore potential be modified tuned achieve customized dynamics, positions them as one most promising hardware candidates executing bioinspired algorithms unique specifications. Finally, identify challenges currently that hinder widespread usage highlight research directions could significantly benefit incorporating IMDs.

Language: Английский

Citations

1